Re: Normalising a two-to-one relationship

Yes, you are right, some do hold that.

I would phrase it somewhat differently. IMO:

Discovering the functional dependencies that are part of the business rules is part of analysis. Very frequently, the best way to discover these dependencies is to start with a proposed or existing schema, and ask questions about how normalized it is.

With this phrasing, normalization is still a feature of the data as designed, rather than of the real world that is being modeled.

But the difference is largely philosophical, not fundamental.

Philip Lijnzaad wrote in message ...
>some hold that normalization is part of the _analysis_,
> not the _design_ . I.e., normalization is something that is
>inherent in the part of real world (or 'business logic') that is being
>modeled.
Received on Tue Jan 02 2001 - 17:36:44 CET